Eppo Inc., a startup that helps software teams test how well new features resonate with users before releasing them, has raised $28 million in Series A funding.

The company detailed today that Innovation Endeavors was the lead investor in the round. Icon VC, Menlo Ventures and Amplify Partners contributed as well. Eppo’s total outside funding now stands at $47.5 million.

Companies find ways of enhancing their websites and applications using a method known as A/B testing. The core idea is to create multiple variations of a feature, such as a button or a product recommendation widget, and pilot each version with a different group of testers. The version that receives the best reception can be rolled out to the application’s entire installed base to improve the user experience. 

Eppo provides a cloud platform that makes it easier to run A/B tests. It also supports a more advanced, AI-powered version of the standard A/B test known as a contextual bandit. The latter experiment format enables software teams to more quickly identify which variation of a new feature will win over the most users. 

In a traditional A/B test, a developer might create three versions of an interface section and test each one with 500 users. The experiment ends once each version has been viewed by the 500 users to whom it was assigned.

In recent years, some companies have switched to a more sophisticated test format known as the multi-armed bandit. Such experiments start similarly to an A/B project: a company creates multiple versions of a feature and pilots each one with a different group of testers. From there, an AI model identifies the version of the feature that performs the best and redirects user traffic to it from the other variations that are being tested. 

According to Eppo, its platform supports an upgraded version of the multi-armed bandit test format known as the contextual bandit. The latter method enhances feature experiments using the information that a company collects about its users. That information, which might include data points such as a customer’s purchase history, helps personalize feature experiments for individual test participants and thereby improves engagement. 

“Experimentation has become the canonical tool of efficient growth, de-risking large investments, ensuring that bad ideas fail fast, and giving leaders high conviction on which initiatives will successfully grow the business,” said Eppo Chief Executive Officer Chetan Sharma.

A/B experiment results often contain a degree of uncertainty that can create accuracy gaps. To address those accuracy issues, companies typically have to collect more data by running additional tests.

Eppo’s platform implements an analytics method called CUPED to reduce the need for additional tests. Using the method, companies can fix accuracy gaps in A/B experiment results with the historical customer data already at their disposal. That removes the need to collect new data by running additional feature experiments and thereby saves time. 

Eppo carries out A/B tests using information in customers’ cloud data warehouses. According to the company, its platform processes much of that information without moving it outside the environment in which it’s stored. This reduces the number of records that customers have to move to Eppo’s infrastructure and thereby lowers the associated network costs.

The company will use the proceeds from its new funding round to hire more workers. In parallel, Eppo plans to extend its platform with features that will allow customers to run more types of A/B tests. The enhancements will focus partly on helping companies find ways of realizing a better return on their marketing budgets. 

Image: Eppo